Hardware-Aware FP4 FlashAttention-4
Blackwell's 4-bit floating-point (FP4) tensor cores do not automatically make attention faster because softmax conversion and on-chip dependencies dominate once its matrix products shrink.
Code that ran · page 2
Papers with a repository link where Syntology ran at least one harvested code sample. Each page shows two streams, newest first within each, counted separately: papers newer than the archive snapshot come from Syntology's graph Syntology; the rest are archive rows archive 2025-07-28. The two are never added together.
Newer than the archive snapshot Syntology
Cards 16–30 of 5,105 graph papers newer than 2025-07-28 with a Syntology-ran sample; this feed shows the newest 150, newest arXiv id first. Dates and the abstract sentence are from arXiv's metadata (CC0) for 5,105 of 5,105.
ENEAS: Embedding-guided Neural Ensemble for Adaptive Segmentation
We present ENEAS, a unified, text-promptable method for instance tracking and semantic discovery.
Fill My Mirror: Geometry-Constrained Mirror Inpainting
Mirrors are common in real-world images, yet producing geometrically consistent reflections with generative models remains challenging.
Understanding Autonomous Driving Datasets by Describing Differences between Image Subsets in Natural Language
Understanding the composition of large-scale autonomous driving datasets is essential for safety, robustness, and reliable operation across domains.
LoRA-TSD: Tangent-Space Spectral Descent for LoRA via Muon-Style Updates
Low-rank adaptation (LoRA) is the standard way to fine-tune large models, yet when its two factors are trained independently, the update ignores the geometry of the low-rank weight change it induces.
Learning to Attract and Repel: Dual Quality Margin Learning for Face Recognition (DQM-Face)
Face recognition in unconstrained environments remains highly challenging due to diverse and extreme variations encountered in real-world scenarios.
Learn from Whoever Is Right: Answer-Verified Multi-Teacher Distillation for Multi-Domain LLMs
Modern large language models (LLMs) rely on reinforcement learning to build strong capabilities in individual domains, but integrating those capabilities into a single deployable model remains challenging.
Rethinking the Teacher-Student Framework for Test-Time Adaptation
Test-Time Adaptation (TTA) has recently emerged as a promising strategy that allows the adaptation of pre-trained models to changing data distributions at deployment time, without access to any labels.
TAME: Temporal-Aware Mixture-of-Experts for Text-Video Retrieval
Text-Video Retrieval (TVR) retrieves videos that match a natural-language query, but extending image-text models such as CLIP to videos is fundamentally limited by the lack of temporal modeling.
Does Playing it Safe Count as Faithfulness? Reassessing LVLM Hallucination Mitigation Methods
Recent inference-time hallucination mitigation methods for large vision-language models (LVLMs) report strong gains on hallucination benchmarks.
Gradient-Update Mismatch: Rethinking Conflict-Free Training of Physics-Informed Neural Networks
Training Physics-Informed Neural Networks (PINNs) requires jointly optimizing physics residual and initial/boundary condition loss terms, which often induce conflicting gradients.
Pix2Rep-v2: Data-Efficient Representation Learning for Dense Medical Imaging Applications
Dense self-supervised learning (SSL) is a powerful paradigm for learning without annotations the local descriptors required to solve dense medical imaging tasks.
Explore More, Drift Less: Outcome-Only Reinforcement Learning Can Suffice for Long-Horizon Interactive Agents
Reinforcement learning is a natural way to post-train LLM agents for long-horizon interactive tasks judged only by end-of-task verification, yet a shared belief holds that outcome-only RL soon hits a ceiling on small…
Dotting the Eye: An Intent-Driven Image Retouching Agent for Visual Focus Enhancement
Image retouching is commonly formulated as enhancing overall visual quality through color adjustment, but in practice, it also serves to emphasize visual focus by guiding viewers' attention toward a specific subject or…
Revisiting Face Recognition for Monozygotic Twins: The Celeb Twins Test Set
Past literature on face recognition for monozygotic (("identical") twins points to facial marks and mirror asymmetry as possible directions for improved accuracy of twins recognition.
From the archive archive 2025-07-28 — filtered to papers Syntology ran Syntology
Cards 16–30 of 31,700 archive papers with a code link, filtered to those where Syntology ran at least one sample (the archive chip labels the rows, the Syntology chip labels the filter; the count is the archive rows that pass it); this feed shows the newest 150, archive date first (newest archive date 2025-07-17). Within a month, dated rows come first, then the 819 undated rows placed by the month in their arXiv id. 1 archive paper with neither a date nor an arXiv id cannot be placed and is not listed.
Scaling RL to Long Videos
We introduce a full-stack framework that scales up reasoning in vision-language models (VLMs) to long videos, leveraging reinforcement learning.
Towards Interpretable Time Series Foundation Models
In this paper, we investigate the distillation of time series reasoning capabilities into small, instruction-tuned language models as a step toward building interpretable time series foundation models.
Open Source Planning & Control System with Language Agents for Autonomous Scientific Discovery
We present a multi-agent system for automation of scientific research tasks, cmbagent.
Addressing Imbalanced Domain-Incremental Learning through Dual-Balance Collaborative Experts
Domain-Incremental Learning (DIL) focuses on continual learning in non-stationary environments, requiring models to adjust to evolving domains while preserving historical knowledge.
SpindleKV: A Novel KV Cache Reduction Method Balancing Both Shallow and Deep Layers
Large Language Models (LLMs) have achieved impressive accomplishments in recent years.
Agent KB: Leveraging Cross-Domain Experience for Agentic Problem Solving
As language agents tackle increasingly complex tasks, they struggle with effective error correction and experience reuse across domains.
Differential Mamba
Sequence models like Transformers and RNNs often overallocate attention to irrelevant context, leading to noisy intermediate representations.
Skywork-R1V3 Technical Report
We introduce Skywork-R1V3, an advanced, open-source vision-language model (VLM) that pioneers a new approach to visual reasoning.
Omni-Video: Democratizing Unified Video Understanding and Generation
Notable breakthroughs in unified understanding and generation modeling have led to remarkable advancements in image understanding, reasoning, production and editing, yet current foundational models predominantly focus…
GTA1: GUI Test-time Scaling Agent
Graphical user interface (GUI) agents autonomously operate across platforms (e.g., Linux) to complete tasks by interacting with visual elements.
pFedMMA: Personalized Federated Fine-Tuning with Multi-Modal Adapter for Vision-Language Models
Vision-Language Models (VLMs) like CLIP have demonstrated remarkable generalization in zero- and few-shot settings, but adapting them efficiently to decentralized, heterogeneous data remains a challenge.
Reviving Cultural Heritage: A Novel Approach for Comprehensive Historical Document Restoration
Historical documents represent an invaluable cultural heritage, yet have undergone significant degradation over time through tears, water erosion, and oxidation.
any4: Learned 4-bit Numeric Representation for LLMs
We present any4, a learned 4-bit weight quantization solution for large language models (LLMs) providing arbitrary numeric representations without requiring pre-processing of weights or activations.
FA: Forced Prompt Learning of Vision-Language Models for Out-of-Distribution Detection
Pre-trained vision-language models (VLMs) have advanced out-of-distribution (OOD) detection recently.
DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge
Recent advances in vision-language-action (VLA) models have shown promise in integrating image generation with action prediction to improve generalization and reasoning in robot manipulation.
The feed is static: 10 pages of up to 15 cards per stream, rebuilt with the site. Older papers are reachable from task, dataset and method pages and from search. No repository stars are tracked and nothing here is ranked by popularity. Machine-readable twin: JSON.